Abstract

Automatic text categorization is a problem of assigning text documents to pre-defined categories. In order to classify text documents, we must extract useful features. In previous researches, a text document is commonly represented by the term frequency and the inverted document frequency of each feature. Since there is a difference between important sentences and unimportant sentences in a document, the features from more important sentences should be considered more than other features. In this paper, we measure the importance of sentences using text summarization techniques. Then we represent a document as a vector of features with different weights according to the importance of each sentence. To verify our new method, we conduct experiments using two language newsgroup data sets: one written by English and the other written by Korean. Four kinds of classifiers are used in our experiments: Naive Bayes, Rocchio, k-NN, and SVM. We observe that our new method makes a significant improvement in all these classifiers and both data sets.

Author Keywords: Text categorization; Importance of sentence; Text summarization techniques; Indexing technique; Text classifier

Article Outline

1. Introduction

2. The proposed text categorization system

2.1. Pre-processing

2.2. Feature selection

2.3. Measuring the importance of sentences

2.3.1. The importance of sentences by the title

2.3.2. The importance of sentences by the importance of terms

2.3.3. The combination of two sentence importance values

2.3.4. The indexing process

3. Other classifiers used in our experiments

3.1. Naive Bayes

3.2. Rocchio

3.3. k-Nearest Neighbor (k-NN) algorithm

3.4. Support vector machines

4. Empirical evaluation

4.1. Data sets and experimental settings

4.2. Experimental results

4.2.1. Setting the number of features

4.2.2. Setting the constant weights k₁ and k₂

4.2.3. Results in two newsgroup data sets

4.3. Verifying our indexing method

5. Conclusions

Acknowledgements

References